unc computational systems biology - lecture01lecture01.ppt author leonard mcmillan created date...

Comp 790-‐087

•  Administra*ve Details

•  Course Overview •  Simple Gene*cs

1/13/14 1 Comp 790– Introduc*on & Administriva

Computational Genetics

Course overview

•  Synopsis –  Graduate-‐level project course –  Guided readings, discussions,

in-‐class exercises, final projects and write-‐up

–  We will probably move to SN325 (will meet here next week)

•  Website –  To appear at:

hUp://www.csbio.unc.edu/mcmillan/index.py?run=Courses.Comp790S14

•  Course Grading –  Class Par*cipa*on 10% –  Research Paper Presenta*on 20% –  Project Proposal 20% –  Final Project & Write-‐up 50%


Syllabus

•  Part 1. Background Topics (1/3) – each student will contribute to a shared s/w library (bring your laptops, preloaded with Python 2.7)

–  Genotyping and Sequencing

–  Recombina*on, phasing, genome mapping

–  Popula*on structure and coalescent theory

–  Selec*on, evolu*on, and phylogene*c trees

–  Epigene*cs, imprin*ng, chroma*n structure, and X inac*va*on

•  Part 2. Paper Presenta*ons (1/3) – each student will be assigned two papers– One as a presenter, the second as a discussant

–  Haplotype assembly from sequence data

–  De novo genome and transcriptome assembly

–  Sequence archival, compara*ve genomic analysis, genomic compression, and query

–  Detec*ng structural varia*on using sequence and genotype data

–  Inferring chroma*n structure using sequence and methyla*on data

•  Part 3. Project Proposals (1/3)

1/13/14 Comp 790– Introduc*on & Administriva 3

Do you want to be genotyped?

•  I have 10 donated kits and will get more if needed

•  Analyze yourself using the tools we develop

•  Contribute your genotypes if you want others to use

•  You must be signed up (not audi*ng) to be genotyped

•  Who in this class is your closest rela*ve?

1/13/14 Comp 665 – Introduc*on & Signals 4

It’s about genes

•  Gene*cs is most clearly understood by considering its subject to be genes rather than organisms

•  Organisms are merely vessels for assuring the survival of genes

•  The central objec*ve of a gene is to propagate itself

•  Successful genes live on long afer their host organism


“[Genes] that survived were the ones that built survival machines for themselves to live in. But making a living got steadily harder as new rivals arose with better and more effective survivial machines. Survival machines got bigger and more eloborate, and the process was cumulative and progressive…”

-- Dawkins, The Selfish Gene

Top-‐down and boBom-‐up

•  Gene*cs results from inheritance

•  Ancestral proper*es can be inferred from extant popula*ons


Exercise

•  Answer the following –  How many dis*nct Y chromosomes are in this

pedigree and who shares them?

–  How many dis*nct Mitochondrial DNAs are in this pedigree and who shares them?

–  Explain how you might reconstruct the paternal X chromosome of siblings B, C, and D?

–  What frac*on of DNA is shared between G and H?

–  What frac*on of DNA is shared by H and I?

–  Suppose that the samples, P, A, D, G, and I have a common phenotype not present in the remainder. Which two samples are most helpful for localizing the gene*c component?


P

C

A

B D E

G

I

F

H

PopulaGon geneGcs

•  Popula*on gene*cs differs from classical gene*cs – Analysis rather than synthesis – Depends on models, which aUempt to explain observa*ons

– Less emphasis on Darwin’s natural selec*on

•  Considers popula*on dynamics –  Isola*on – BoUlenecks


Why computer science

•  Classically, gene*cs, both top-‐down (classical) and boUom-‐up (popula*on) have focused on mathema*cal/sta*s*cal models

•  Access to gene*cs data has recently outpaced our capability to process, interpret, and analyze it

•  As model complexity increases, it becomes harder to find closed-‐form solu*ons

•  Relies more and more on computa*onal modeling to infer structure, account for noise etc.


Wright-‐Fisher model

•  One of the first, and simplest models of popula*on genealogies was introduced by Wright (1931) and Fisher (1930).

•  Model emphasizes transmission of genes from one genera*on to the next

•  For simplicity we’ll first focus on a fixed popula*on size, each with a dis*nct gene variant


Simple haploid model

•  Rules –  Antecedent genes are chosen randomly, with replacement, from their parental genera*on

–  No selec*on –  Fixed popula*on size


G0: ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J']

G1: ['J', 'A', 'H', 'B', 'I', 'E', 'D', 'G', 'A', 'B']

G2: ['A', 'J', 'E', 'G', 'D', 'E', 'B', 'I', 'A', 'A']

G3: ['A', 'A', 'E', 'J', 'I', 'A', 'I', 'A', 'J', 'B']

G4: ['E', 'A', 'B', 'B', 'A', 'E', 'A', 'A', 'A', 'A']

G5: ['A', 'A', 'B', 'A', 'A', 'E', 'A', 'A', 'A', 'B']

G6: ['A', 'A', 'A', 'A', 'A', 'A', 'A', 'B', 'A', 'A']

G7: ['B', 'A', 'A', 'B', 'A', 'A', 'A', 'A', 'A', 'A']

What will this population eventually look like?

AssumpGons of Wright/Fisher

•  Discrete and non-‐overlapping genera*ons •  Haploid individuals •  Popula*ons size is constant •  All individuals are equally fit •  No popula*on of social structure •  Genes segregate independently


Some Graphical AbstracGons

•  Replace leUers with colors

•  Draw lineages •  Sort topologically


Repeats


Every population results in just one gene

Onset of uniformity

•  10000 trials •  Mode = 11 (616)

•  Mean = 17.5


Next Time

•  Gene*cs Background –  Inbreeding –  Coefficient of relatedness

–  Diploidy –  Recombina*on

–  Phasing •  Bring your laptops loaded with Python 2.7

– We will analyze real genotypes

•  A list of recent papers to choose from


unc computational systems biology - lecture01lecture01.ppt author leonard mcmillan created date...

Documents